| In the past decades,the development of lidar(light detection and ranging)technology provides a new method to obtain high-resolution geospatial information.Airborne lidar system can acquire three-dimensional(3D)information quickly and accurately by actively emitting laser pulses to the ground.At present,lidar data has been widely used in many fields,such as digital elevation model(DEM)acquisition,road extraction and forest structure parameter retrieval.In most airborne lidar applications,the key step is point cloud filtering,which is the process of removing non ground points and retaining ground points.The filtering result directly affects the quality of DEM,which is one of the key factors affecting the subsequent processing.Therefore,how to quickly,effectively and accurately obtain the ground points has been a hot issue for many scholars.Aiming at the hot issue of point cloud filtering,this paper mainly does the following work;1.Survey of point cloud filtering algorithm.According to different principles and theoretical background,six kinds of filtering algorithms,including slope based,mathematical morphology based,surface based,segmentation based,machine learning / deep learning based and hybrid filtering algorithm,are described in detail and systematically,and the advantages,disadvantages and adaptability of various filtering algorithms are comprehensively evaluated and analyzed.2.Adaptability analysis of mathematical morphology and cloth simulation filteringalgorithm.The point cloud filtering algorithm based on mathematical morphology has the characteristics of simple principle and high efficiency,so it has been studied by many scholars,and the related algorithms have been improved.In recent years,the filtering algorithm of cloth simulation has been widely concerned because it has few user-defined parameters and is easy to set.In this paper,15 groups of experimental samples are selected for filtering experiments and filtering accuracy evaluation of the two algorithms,and adaptability analysis is carried out.It provides more references and suggestions for the improvement of the correlation filtering algorithm.3.Research on improved cloth simulation filtering algorithm combined withprogressive morphology.When cloth simulation filtering algorithm is used to deal with super large and low buildings,cloth particles may stick to the roof,and some non-ground points may be mistakenly classified as ground points,which will affect the filtering effect.In this paper,a cloth simulation filtering algorithm combined with progressive morphology is proposed.A small morphological filtering window is selected for primary filtering to obtain the initial ground point cloud.On the basis of the initial ground point cloud,cloth simulation filtering algorithm is used for secondary filtering to obtain the final ground point cloud.The experimental results show that the proposed algorithm can effectively reduce the class I error and total error of the city and other complex terrain areas,and effectively retain the terrain details while filtering the surface features.4.Ground feature extraction based on filtering algorithm.Ground micro topography plays an important role in the discovery and research of ancient archaeological sites.However,there are few researches on using filtering algorithm to extract ground features and retaining ground micro topography.In this paper,by comparing four classical point cloud filtering algorithms,five groups of specific terrain are selected,and some specific features are marked to carry out feature extraction experiments,and the results of feature extraction are analyzed qualitatively and quantitatively.This paper compares and analyzes the adaptability of four filtering algorithms in ground feature extraction,which provides a certain reference for the practical application of filtering algorithm in ground feature extraction. |
PDF Full Text Request